Precision direct power control device for dual active bridge converters

Abstract

This application provides a precise direct power control device for a dual active bridge converter, relating to the field of power control technology. The device includes: a signal processing module for acquiring real-time electrical signal sequences and performing sampling interference correction; a power calculation module for constructing a dual active bridge power calculation model and obtaining instantaneous transmitted power by inputting a reliable electrical signal sequence; a feedback adjustment module for determining the power error signal and performing direct feedback adjustment; and a parameter determination module for dynamically optimizing the dual active bridge adjustment candidate space based on the dynamic accompanying risk mechanism of the dual active bridge to obtain the dual active bridge adjustment parameters. This application solves the technical problem in the prior art where power control only focuses on power tracking during dynamic adjustment and ignores transient safety risks, leading to potential safety hazards during load changes and thus affecting the operating efficiency of the dual active bridge converter. It effectively suppresses dynamic risks while achieving precise power control, thereby improving the operating efficiency of the dual active bridge converter.

Description

Technical Field

[0001] This application relates to the field of power control technology, and more specifically to a precise direct power control device for a dual active bridge converter. Background Technology

[0002] Existing direct power control methods for dual active bridge converters often focus solely on the speed of power point tracking during dynamic adjustment, neglecting the transient safety risks that may arise during the adjustment process itself. Especially during sudden load changes or power command jumps, these changes can easily lead to an imbalance in the volt-second product across the transformer, resulting in transient DC bias currents. This can cause safety hazards such as inductor current spikes and unidirectional saturation of the transformer core, and in severe cases, even threaten the safe operation of switching devices and the transformer. Consequently, the converter may become unstable during load changes, reducing the operating efficiency of the dual active bridge converter.

[0003] In summary, existing technologies suffer from the problem that power control focuses only on power tracking during dynamic adjustment while ignoring transient safety risks, which can lead to safety hazards during load surges and consequently affect the operating efficiency of dual active bridge converters. Summary of the Invention

[0004] The purpose of this application is to provide a precise direct power control device for a dual active bridge converter, in order to solve the technical problem in the prior art that, due to the fact that the power control only focuses on power tracking and ignores transient safety risks during the dynamic adjustment process, safety hazards are easily generated when the load changes suddenly, which in turn affects the operating efficiency of the dual active bridge converter.

[0005] To achieve the above objectives, this application provides a precise direct power control device for a dual active bridge converter, comprising: a signal processing module for acquiring a real-time electrical signal sequence of the dual active bridge converter and performing sampling interference correction on the real-time electrical signal sequence to obtain a reliable electrical signal sequence; a power calculation module for constructing a dual active bridge power calculation model and inputting the reliable electrical signal sequence into the dual active bridge power calculation model to obtain instantaneous transmission power; a feedback adjustment module for comparing the instantaneous transmission power with a target reference power to determine a power error signal, and performing direct feedback adjustment on the dual active bridge converter according to the dual active bridge power calculation model and the power error signal to obtain a dual active bridge adjustment candidate space that satisfies the power error tolerance range; and a parameter determination module for performing dynamic risk optimization on the dual active bridge adjustment candidate space according to the dual active bridge dynamic accompanying risk mechanism to obtain dual active bridge adjustment parameters.

[0006] Optionally, the signal processing module includes: a sampling scene parameter acquisition unit, used to acquire sampling scene parameters of the real-time electrical signal sequence to obtain a sampling scene dataset; an interference recognition unit, used to perform interference recognition based on the sampling scene dataset to determine the sampling scene interference characteristic set; an interpolation reconstruction unit, used to perform interpolation reconstruction on the real-time electrical signal sequence based on the sampling scene interference characteristic set to obtain a reconstructed electrical signal sequence; and a signal sequence processing unit, used to perform digital low-pass filtering and timestamp alignment correction on the reconstructed electrical signal sequence to generate the reliable electrical signal sequence.

[0007] Optionally, the dual active bridge converter includes a primary-side full bridge, a secondary-side full bridge, and an isolation transformer.

[0008] Optionally, the power calculation module includes the function expression of the dual active bridge power calculation model as follows: Where P is the instantaneous transmission power, V1 is the primary DC voltage, V2 is the secondary DC voltage, f is the switching frequency, L is the total equivalent inductance on both sides of the isolation transformer, d is the outward phase shift angle, d1 is the primary inward phase shift angle, d2 is the secondary inward phase shift angle, and n is the harmonic order.

[0009] Optionally, the feedback adjustment module includes: a signal judgment unit, used to judge whether the power error signal meets the power error tolerance range; and an instruction generation unit, used to generate a steady-state power maintenance instruction if the power error signal meets the power error tolerance range.

[0010] Optionally, the feedback adjustment module includes: a control record retrieval unit, used to retrieve control records of the dual active bridge converter according to the target reference power if the power error signal does not meet the power error tolerance range, and obtain a power matching control sample set; a statistical clustering unit, used to perform statistical clustering according to the power matching control sample set, and determine multiple adjustment variable center points and multiple variable cluster dispersions; a boundary determination unit, used to construct multiple variable parameter boundaries according to the multiple adjustment variable center points and the multiple variable cluster dispersions; a parameter feedback adjustment unit, used to perform feedback adjustment of the dual active bridge converter according to the multiple variable parameter boundaries, and obtain the dual active bridge feedback adjustment space; and a parameter optimization identification unit, used to perform optimization identification of the dual active bridge feedback adjustment space according to the dual active bridge power calculation model and the power error tolerance range, and establish the dual active bridge adjustment candidate space.

[0011] Optionally, the parameter optimization and identification unit includes: a dynamic modeling subunit, used to perform dynamic modeling based on the dual active bridge converter to obtain a dual active bridge converter model; a simulation adjustment subunit, used to simulate and adjust the dual active bridge converter model according to each dual active bridge feedback adjustment scheme within the dual active bridge feedback adjustment space to obtain multiple scheme simulation data; a scheme simulation subunit, used to input the multiple scheme simulation data into the dual active bridge power calculation model to obtain multiple transmission power simulation parameters; a parameter discrimination subunit, used to determine whether the multiple transmission power simulation parameters meet the power error tolerance range to obtain multiple power simulation verification results; and a result filtering subunit, used to optimize and filter the dual active bridge feedback adjustment space based on the multiple power simulation verification results to obtain the dual active bridge adjustment candidate space.

[0012] Optionally, the parameter determination module includes: a risk mechanism activation unit, used to activate the dual active bridge dynamic accompanying risk mechanism, the dual active bridge dynamic accompanying risk mechanism including a dual active bridge accompanying risk analysis layer, a dual active bridge accompanying risk constraint layer, and a dual active bridge accompanying risk weight layer; an accompanying risk constraint optimization unit, used to perform accompanying risk constraint optimization on the dual active bridge adjustment candidate space according to the dual active bridge accompanying risk analysis layer and the dual active bridge accompanying risk constraint layer, to obtain a dual active bridge adjustment optimization space; a risk weighted calculation unit, used to perform risk weighted calculation on the dual active bridge adjustment optimization space according to the dual active bridge accompanying risk weight layer, to obtain a dynamic risk cost distribution; and a dynamic risk cost iterative optimization unit, used to perform dynamic risk cost iterative optimization on the dual active bridge adjustment optimization space according to the dynamic risk cost distribution, to generate the dual active bridge adjustment parameters.

[0013] Optionally, the accompanying risk constraint optimization unit includes: an accompanying risk prediction subunit, used to predict the accompanying risks of the dual active bridge regulation candidate space according to the dual active bridge accompanying risk analysis layer, and obtain multiple accompanying risk sequences, wherein the accompanying risk analysis indicators of the dual active bridge accompanying risk analysis layer include power reverse mutation risk, overshoot oscillation risk, and equipment failure risk; a risk verification result generation subunit, used to input the multiple accompanying risk sequences into the dual active bridge accompanying risk constraint layer, and obtain multiple accompanying risk verification results, wherein the dual active bridge accompanying risk constraint layer includes power reverse mutation risk constraint, overshoot oscillation risk constraint, and equipment failure risk constraint; and a risk verification result filtering subunit, used to filter the dual active bridge regulation candidate space according to the multiple accompanying risk verification results, and establish the dual active bridge regulation optimization space.

[0014] Optionally, the dual active bridge accompanying risk weight layer includes power reverse mutation risk weight, overshoot oscillation risk weight, and equipment failure risk weight.

[0015] The technical solution provided in this application has at least the following technical effects or advantages:

[0016] The signal processing module acquires the real-time electrical signal sequence of the dual active bridge converter and performs sampling interference correction on the real-time electrical signal sequence to obtain a reliable electrical signal sequence. The power calculation module constructs a dual active bridge power calculation model and inputs the reliable electrical signal sequence into the dual active bridge power calculation model to obtain the instantaneous transmission power. The feedback adjustment module compares the instantaneous transmission power with the target reference power to determine the power error signal, and performs direct feedback adjustment on the dual active bridge converter according to the dual active bridge power calculation model and the power error signal to obtain a dual active bridge adjustment candidate space that meets the power error tolerance range. The parameter determination module performs dynamic risk optimization on the dual active bridge adjustment candidate space according to the dual active bridge dynamic accompanying risk mechanism to obtain the dual active bridge adjustment parameters. In other words, by acquiring the real-time electrical signal sequence of the dual active bridge converter and performing interference correction, the instantaneous transmission power is calculated in real time according to the power calculation model. The instantaneous transmission power is compared with the target reference power, and the power error signal is obtained for feedback adjustment. A dynamic risk mechanism of dual active bridge is introduced to dynamically optimize the adjustment candidate space. The adjustment parameters of the converter are optimized according to the real-time risk assessment. While achieving precise power control, dynamic risks are effectively suppressed, and the operating efficiency of the dual active bridge converter is significantly improved.

[0017] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the precise direct power control device for the dual active bridge converter of this application.

[0020] Figure 2This is a schematic diagram of the parameter determination module in the precise direct power control device of the dual active bridge converter of this application.

[0021] Explanation of reference numerals in the attached figures: Signal processing module 11, power calculation module 12, feedback adjustment module 13, parameter determination module 14, risk mechanism activation unit 41, accompanying risk constraint optimization unit 42, risk weighted calculation unit 43, dynamic risk cost iterative optimization unit 44. Detailed Implementation

[0022] This application provides a precise direct power control device for a dual active bridge converter, solving the technical problem in existing technologies where power control focuses only on power tracking during dynamic adjustment, neglecting transient safety risks. This leads to safety hazards during load surges, thus affecting the operating efficiency of the dual active bridge converter. By acquiring the real-time electrical signal sequence of the dual active bridge converter and performing interference correction, the instantaneous transmitted power is calculated in real-time according to the power calculation model. The instantaneous transmitted power is compared with the target reference power to obtain a power error signal, which is then used for feedback adjustment. A dynamic risk mechanism associated with the dual active bridge is introduced to dynamically optimize the adjustment candidate space. Based on real-time risk assessment, the converter's adjustment parameters are optimized, achieving precise power control while effectively suppressing dynamic risks and significantly improving the operating efficiency of the dual active bridge converter.

[0023] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0024] For examples, please refer to the appendix. Figure 1 This application provides a precise direct power control device for a dual active bridge converter, wherein the precise direct power control device for the dual active bridge converter includes:

[0025] The signal processing module 11 is used to acquire the real-time electrical signal sequence of the dual active bridge converter and perform sampling interference correction on the real-time electrical signal sequence to obtain a reliable electrical signal sequence.

[0026] Furthermore, the signal processing module 11 in the precise direct power control device of the dual active bridge converter is also used for: a sampling scene parameter acquisition unit, used to acquire sampling scene parameters of the real-time electrical signal sequence to obtain a sampling scene dataset; an interference recognition unit, used to perform interference recognition based on the sampling scene dataset to determine the sampling scene interference characteristic set; an interpolation reconstruction unit, used to perform interpolation reconstruction on the real-time electrical signal sequence based on the sampling scene interference characteristic set to obtain a reconstructed electrical signal sequence; and a signal sequence processing unit, used to perform digital low-pass filtering and timestamp alignment correction on the reconstructed electrical signal sequence to generate the reliable electrical signal sequence.

[0027] Furthermore, the signal processing module 11 in the precise direct power control device of the dual active bridge converter is also used for: the dual active bridge converter includes a primary-side full bridge, a secondary-side full bridge, and an isolation transformer.

[0028] Specifically, a dual active bridge converter is a power electronic device typically used for efficient power conversion and transmission, comprising a primary-side full-bridge, a secondary-side full-bridge, and an isolation transformer. The primary-side full-bridge is an H-bridge circuit composed of four switching transistors (such as metal-oxide-semiconductor field-effect transistors or insulated-gate bipolar transistors), connected to the input DC source side, responsible for converting DC to a high-frequency AC square wave. The secondary-side full-bridge is also an H-bridge circuit composed of four switching transistors, connected to the output load side, responsible for re-rectifying the high-frequency AC square wave back to DC. The isolation transformer is a high-frequency transformer connected between the primary and secondary full-bridges, providing electrical isolation and voltage matching, and simultaneously forming an energy storage element for energy transmission through its leakage inductance and the applied inductance.

[0029] Voltage and current sensors installed at key locations in the main circuit of the dual active bridge converter periodically read analog outputs from voltage Hall effect sensors and current Hall effect sensors at a fixed high sampling frequency. These sensors are installed at both ends of the primary DC bus, both ends of the secondary DC bus, and on both sides of the series current sensing resistor or current transformer on the primary side of the transformer. Each sampling yields a set of values, such as the instantaneous values ​​of the primary voltage, primary current, and secondary voltage at the current moment. These continuously generated sampling points are stored in the controller's data buffer in chronological order, thus forming a real-time electrical signal sequence.

[0030] While sampling electrical signals, or within a very short time, the control system simultaneously acquires a series of parameters reflecting the current operating scenario, including the current output power level of the converter, the on-temperature of the switching transistors, the converter's operating mode, and the electromagnetic noise floor. These sampling scenario parameters, at the moment of signal sampling, provide a series of auxiliary information describing the converter's current operating state and environmental conditions, helping to determine the reliability of the sampled signal and the types of interference it may be subject to.

[0031] The collected sampling scene dataset is input into a built-in interference recognition rule base. This rule base is built based on prior knowledge and a large amount of experimental data. A logical example of the rule base is as follows: if the output power is greater than 2500W and the temperature is higher than 50℃, it is determined to be a high-power, high-temperature scene. In this case, the electromagnetic interference generated by the switching process of the switching transistor is strong, and the main interference type is high-frequency common-mode noise. If the amplitude of multiple consecutive jumps in the sampled value exceeds the normal range, it is determined to be pulse spike interference. If the phase difference between the voltage and current sampling channels deviates too much from the theoretical value, it is determined to be sampling delay mismatch. Based on these rules, the interference characteristics of the current sample are identified, forming a feature set containing one or more interference types. Interference recognition is the process of determining what type of interference or error might affect the current sampled signal based on the collected scene dataset, using preset rules or lookup tables. Through interference recognition technology, it is analyzed which parts of the data are valid signals and which parts are interference signals. The sampling scene interference feature set identifies the specific type of interference and its severity in the current sampled data, whether it is high-frequency switching noise, sampling delay, or pulse spike interference.

[0032] Based on the interference characteristic set of the sampling scene, the interpolation reconstruction module is activated, traversing the real-time electrical signal sequence. According to the interference type and occurrence time identified in the interference characteristic set, it marks which sampling points are contaminated. For example, for pulse interference, isolated points with amplitudes far exceeding the average of adjacent points are identified; for high-frequency noise, point-by-point replacement is not necessarily performed. For each marked contaminated point, the values ​​of several uncontaminated sampling points before and after that point are extracted, such as the first three points and the last three points. Based on the sampling time interval and the location of the contaminated point, an appropriate interpolation algorithm is selected for calculation. Linear interpolation can be used for slowly changing signals, and third-order spline interpolation can be used for rapidly changing signals. The calculated new values ​​will replace the outliers in the original sampling points, and all points together constitute the reconstructed electrical signal sequence. For example, suppose the primary voltage sampling values ​​in the original real-time electrical signal sequence are…, 399.8, 400.1, 415.0, 400.2, 399.9,… Based on the interference characteristic set (including slight pulse interference), the third point 415.0 was identified as a contaminated point due to excessively large jumps compared to the points before and after it, indicating pulse interference. The values ​​of the two points before this point (399.8 and 400.1) and the two points after it (400.2 and 399.9) were extracted, and a second-order linear interpolation algorithm was used for reconstruction. The calculation process was as follows: assuming the true value should closely approximate the trend of the preceding and following points, the average of the preceding point 400.1 and the following point 400.2 was taken, and then slightly adjusted to obtain the reconstructed value 400.15V. This 400.15V was used to replace the original 415.0V. After performing similar processing on all marked points, the reconstructed electrical signal sequence was obtained.

[0033] The reconstructed electrical signal sequence is fed into a digital low-pass filter module. The filter's cutoff frequency is typically designed to be approximately twice the converter's switching frequency; for example, for a 100kHz switching frequency, the cutoff frequency is set to 200kHz. Each sampling point in the sequence participates in the filtering operation sequentially, and the filter outputs a smoothed new sequence, significantly suppressing high-frequency glitches and residual noise. The filtered sequence then enters the timestamp alignment correction module. Because the propagation delays of the primary-side voltage sampling circuit and the primary-side current sampling circuit may differ (e.g., a 0.5µs delay for the voltage sensor and a 1µs delay for the current sensor), the voltage and current sampled at the same moment may actually correspond to different physical times. Based on the pre-measured or calibrated delay times of each channel, the current sequence is time-shifted. For example, the entire current sequence is shifted forward or backward by 0.5µs to ensure its timestamps are strictly aligned with the voltage sequence. After these two steps, the final generated data sequence is a reliable electrical signal sequence. Sampling interference correction using the real-time electrical signal sequence significantly improves the accuracy and stability of the dual active bridge converter power control system.

[0034] The power calculation module 12 is used to construct a dual active bridge power calculation model and input the reliable electrical signal sequence into the dual active bridge power calculation model to obtain the instantaneous transmission power.

[0035] Furthermore, the power calculation module 12 in the precise direct power control device of the dual active bridge converter is also used for: the functional expression of the dual active bridge power calculation model is: Where P is the instantaneous transmission power, V1 is the primary DC voltage, V2 is the secondary DC voltage, f is the switching frequency, L is the total equivalent inductance on both sides of the isolation transformer, d is the outward phase shift angle, d1 is the primary inward phase shift angle, d2 is the secondary inward phase shift angle, and n is the harmonic order.

[0036] Specifically, the dual active bridge power calculation model is a mathematical expression based on Fourier series expansion, used to accurately calculate the instantaneous active power transmitted from the primary side to the secondary side of the dual active bridge converter at any given three phase shift angles: the outer phase shift angle, the primary side inner phase shift angle, and the secondary side inner phase shift angle. Unlike traditional simplified models, this model not only considers the contribution of the fundamental component but also incorporates the effects of all higher harmonics, thus maintaining high accuracy across the entire operating range.

[0037] The functional expression for the dual active bridge power calculation model is: Where, P is the instantaneous transmitted power, the instantaneous active power value actually transmitted by the converter during the current switching cycle; V1 is the primary side DC voltage, i.e., the average or instantaneous voltage value of the DC input side of the primary full-bridge converter; V2 is the secondary side DC voltage, i.e., the average or instantaneous voltage value of the DC output side of the secondary full-bridge converter; f is the switching frequency, i.e., the switching frequency of the power switching transistors of the converter, which determines the basic cycle of power transmission; L is the total equivalent inductance resistance on both sides of the isolation transformer, i.e., the sum of the external inductance value connected in series on the primary or secondary side of the isolation transformer and the leakage inductance value referred to one side of the transformer; d is the external phase shift. The phase difference between the AC square wave voltage output from the primary side of the full-bridge and the AC square wave voltage output from the secondary side of the full-bridge is d1, which is the phase difference between the switching transistor drive signals of the two bridge arms inside the primary side of the full-bridge, controlling the duty cycle and waveform of the output voltage of the primary side of the full-bridge; the phase difference between the switching transistor drive signals of the two bridge arms inside the secondary side of the full-bridge is d2, which is the phase difference between the switching transistor drive signals of the two bridge arms inside the secondary side of the full-bridge, controlling the duty cycle and waveform of the output voltage of the secondary side of the full-bridge; and the harmonic order is the multiple of each harmonic relative to the fundamental frequency in the Fourier series expansion, which approximates the true power waveform by accumulating the contributions of infinitely many harmonics.

[0038] When calculating the instantaneous transmitted power at the current moment, the primary and secondary DC voltage values ​​are first extracted from the reliable electrical signal sequence. Simultaneously, the values ​​of the currently executing external phase shift angle, primary-side internal phase shift angle, and secondary-side internal phase shift angle are read from the controller's pulse width modulation register. The switching frequency and equivalent inductance, as fixed parameters of the converter, are determined and stored in memory during system initialization. Using these seven parameters as input, a pre-written dual active bridge power calculation model is invoked to calculate the fundamental component. The fundamental component is the first term in the formula, i.e., the linear term of the external phase shift angle d multiplied by a coefficient, which is obtained by dividing the product of the primary and secondary voltages by 2π, multiplying by the switching frequency f, and multiplying by the equivalent inductance. The harmonic correction term corresponds to the summation part in the formula. The controller sets an accumulation loop, starting from harmonic order n equal to 1, successively calculating the value of each term and accumulating them. The calculation of each term includes: calculating 1 / n 2 The calculation involves determining the cosine value of nπd, the sine value of nπd1, and the cosine value of nπd2. These four parts are then multiplied together and multiplied by -2 / π. Since theoretically this is an infinite series, in practical engineering implementation, the controller is set to a maximum harmonic order, such as calculating up to n equal to the 50th or 100th harmonic. Accumulation stops when the values ​​of subsequent terms become negligible. The fundamental component and the harmonic correction terms are then summed to obtain the final instantaneous transmitted power value. The entire calculation process is completed within one control cycle (e.g., ten microseconds), ultimately outputting an instantaneous transmitted power value for subsequent error comparison and feedback adjustment.

[0039] The instantaneous power of the dual active bridge converter is calculated by using a dual active bridge power calculation model, which accurately reflects the influence of electrical parameters and phase angle in the converter on power transmission, thereby helping to adjust the working state of the converter and achieve precise power control.

[0040] The feedback adjustment module 13 is used to compare the instantaneous transmission power with the target reference power, determine the power error signal, and perform direct feedback adjustment on the dual active bridge converter according to the dual active bridge power calculation model and the power error signal to obtain the dual active bridge adjustment candidate space that meets the power error tolerance range.

[0041] Furthermore, the feedback adjustment module 13 in the precise direct power control device of the dual active bridge converter is also used for: a signal judgment unit, used to judge whether the power error signal meets the power error tolerance range; and an instruction generation unit, used to generate a steady-state power maintenance instruction if the power error signal meets the power error tolerance range.

[0042] Specifically, the power error is obtained by comparing the instantaneous transmitted power with the target reference power and performing a simple subtraction operation; that is, the difference between the two is calculated to obtain the power error signal. The target reference power is a set ideal power value or reference value, usually set according to external conditions such as load demand and input voltage, serving as the target for system operation. The power error signal refers to the difference between the instantaneous transmitted power and the target reference power, reflecting the deviation between the current actual output power and the expected target power, and is an important basis for regulation and control.

[0043] The power error tolerance range is the allowable deviation range of the power error signal. Therefore, a tolerance error range is set according to the control accuracy and safety requirements of the dual active bridge converter. If the error exceeds this range, the dual active bridge converter is considered to have become unstable or abnormal.

[0044] The system determines whether the power error signal meets the power error tolerance range. If the power error signal meets the tolerance range (i.e., less than the upper limit and greater than the lower limit), the output power of the dual active bridge converter is considered stable and normal. At this point, a steady-state power sustaining command is generated, instructing the converter to maintain the current power state and avoid further frequent adjustments, thus maintaining stability. The sustaining command typically keeps the converter at its current operating point, preventing power output fluctuations or overshoot. In some cases, this also means the converter will enter a standby state, waiting for load changes or input voltage fluctuations before adjusting. When the converter's power output approaches the target power, a steady-state power sustaining command is automatically generated to maintain system stability and avoid unnecessary frequent adjustments.

[0045] Furthermore, the feedback adjustment module 13 in the precise direct power control device of the dual active bridge converter is also used for: a control record retrieval unit, used to perform control record retrieval on the dual active bridge converter according to the target reference power if the power error signal does not meet the power error tolerance range, and obtain a power matching control sample set; a statistical clustering unit, used to perform statistical clustering according to the power matching control sample set, and determine multiple adjustment variable center points and multiple variable cluster dispersion; a boundary determination unit, used to construct multiple variable parameter boundaries according to the multiple adjustment variable center points and the multiple variable cluster dispersion; a parameter feedback adjustment unit, used to perform feedback adjustment on the dual active bridge converter according to the multiple variable parameter boundaries, and obtain the dual active bridge feedback adjustment space; and a parameter optimization identification unit, used to perform optimization identification on the dual active bridge feedback adjustment space according to the dual active bridge power calculation model and the power error tolerance range, and establish the dual active bridge adjustment candidate space.

[0046] Furthermore, the feedback adjustment module 13 in the precise direct power control device of the dual active bridge converter is also used for: a dynamic modeling subunit, used for dynamically modeling the dual active bridge converter to obtain a dual active bridge conversion model; a simulation adjustment subunit, used for simulating and adjusting the dual active bridge conversion model according to each dual active bridge feedback adjustment scheme in the dual active bridge feedback adjustment space to obtain multiple scheme simulation data; a scheme simulation subunit, used for inputting the multiple scheme simulation data into the dual active bridge power calculation model to obtain multiple transmission power simulation parameters; a parameter discrimination subunit, used for judging whether the multiple transmission power simulation parameters meet the power error tolerance range to obtain multiple power simulation test results; and a result filtering subunit, used for optimizing and filtering the dual active bridge feedback adjustment space according to the multiple power simulation test results to obtain the dual active bridge adjustment candidate space.

[0047] Specifically, if the power error signal does not meet the power error tolerance range (i.e., it is less than the minimum value or greater than the maximum value), it indicates a deviation in the output power, requiring further adjustment. Control records are retrieved based on the current target reference power, searching historical control data for the control sample closest to the current target power. The aim is to find the optimal adjustment scheme for similar system performance. Control records typically include information such as power output, control parameters, and dual active bridge converter status during past adjustments. The power matching control sample set refers to the set of control samples in historical data that best match the current target reference power, providing records of different control parameters and adjustment schemes. If the number of retrieved samples is insufficient, the sample set can be expanded by adjusting the matching threshold, such as widening the voltage deviation range, or by waiting for the next control cycle to accumulate more data.

[0048] Statistical clustering is performed on the power matching control sample set, grouping multiple control samples according to similarity. Based on historical sample data, the centroid and dispersion of each control variable are determined. A sample point is randomly selected as the initial centroid, and then the mean of all points within a certain range around it is calculated, with this mean used as the new centroid. This process is iterated until the centroids converge. In this way, the algorithm can identify several major clustering regions in the sample set. After clustering, each clustering region corresponds to a category. The average of the three phase shift angles of all sample points within that category is calculated to obtain the centroid of the control variable for that category. Simultaneously, the covariance matrix or standard deviation of each dimension of that category is calculated to obtain the variable cluster dispersion. A larger dispersion value indicates a wider range of control parameter variation for that category; a smaller dispersion value indicates a highly concentrated control parameter distribution. The centroid of the control variable refers to the location of the center point after data clustering, representing the optimal combination of control parameters in each cluster group, usually the average or optimal configuration of the variable; the cluster dispersion refers to the range of sample point distribution, reflecting the fluctuation range of each control variable. If the dispersion of a cluster is large, it indicates that the differences between samples are greater, suggesting that the parameters of that cluster change more drastically and require a more refined adjustment scheme.

[0049] Based on the centroids of multiple control variables and the dispersion of multiple variable clusters, a parameter boundary is constructed for each control variable, i.e., the effective operating range of the variable. These parameter boundaries ensure that each control variable varies within its optimal range, thereby maintaining the stability of the dual active bridge converter and avoiding instability caused by over-adjustment of parameters. The parameter boundary, based on cluster analysis, determines the effective parameter range of each control variable, ensuring that each control variable can be adjusted within its effective range to guarantee the stability and efficiency of the dual active bridge converter.

[0050] Based on multiple variable parameter boundaries, real-time feedback adjustment is performed on the dual active bridge converter to obtain the dual active bridge feedback adjustment space, which includes multiple dual active bridge feedback adjustment schemes. These schemes are used for real-time adjustment when the power error exceeds the allowable range to ensure accurate power transmission. A grid scanning method is used, sampling the three phase shift angles at equal intervals within each boundary according to a pre-set step size. For example, for the first type of boundary, the outer phase shift angle step size is set to 0.005 radians, the primary side inner phase shift angle step size is 0.003 radians, and the secondary side inner phase shift angle step size is 0.003 radians. Then, starting from the lower limit of the boundary, a series of grid points are generated with increasing step sizes. Each grid point corresponds to a complete combination of three phase shift angles. Listing all these grid points constitutes a set of discrete adjustment schemes. The same grid scanning operation is performed for each variable parameter boundary, and then the adjustment schemes generated for all boundaries are merged together to form a larger set, which is the dual active bridge feedback adjustment space. Each element in this space, that is, each set of phase shift angles, constitutes a specific dual active bridge feedback regulation scheme.

[0051] A high-fidelity mathematical model is established based on the actual hardware parameters of the dual active bridge converter. This model is constructed using the state-space averaging method or the switching cycle averaging method, with its core being a set of differential equations describing the time-varying inductor current and capacitor voltage. In other words, based on the physical topology and component parameters of the dual active bridge converter, a dual active bridge converter model is built in the controller software to simulate the converter's dynamic behavior, reflecting its transient response under different input conditions and control parameters. The dual active bridge converter model is a software module that encapsulates the converter's state equations or transfer function, capable of receiving control inputs such as phase shift angles and outputting corresponding electrical quantities such as voltage, current, and power.

[0052] The input variables of the dual active bridge converter model include the primary voltage, secondary voltage, and three phase shift angles (outer phase shift angle, primary inner phase shift angle, and secondary inner phase shift angle). The internal parameters of the dual active bridge converter model include the transformer equivalent inductance, the DC bus capacitance values ​​of the primary and secondary sides, the on-resistance and voltage drop of the switching transistors, etc., all of which are obtained from the actual hardware design. The output variables of the dual active bridge converter model include the inductor current waveform, the instantaneous value of the primary current, the instantaneous value of the secondary voltage, and the instantaneous transmitted power. Each time a simulation adjustment is required, the dual active bridge converter model is invoked, using the current primary and secondary voltages as the initial conditions of the model, and the combination of phase shift angles to be tested as the input excitation. The differential equations are solved using numerical integration methods, thereby simulating the complete dynamic process of the dual active bridge converter transitioning from the current state to a new steady state under given control parameters.

[0053] From the dual-active-bridge feedback regulation space, each dual-active-bridge feedback regulation scheme is taken out in sequence, that is, each combination of external phase-shift angle, primary-side internal phase-shift angle, and secondary-side internal phase-shift angle. For each scheme, set the actual values of the primary-side voltage and the secondary-side voltage at the current moment as the initial state of the transformation model; set the currently executing phase-shift angle combination as the initial control parameter of the model, and set the new phase-shift angle combination to be tested as the target control parameter of the model. Simulate that the controller switches the phase-shift angle from the old value to the new value at a certain moment; run the dual-active-bridge transformation model, and calculate recursively forward with a time step of microseconds to simulate the change trajectories of the inductor current, primary-side voltage, and secondary-side voltage in the next few hundred microseconds or even a few milliseconds starting from the switching moment; after the simulation time reaches the preset adjustment time, record the steady-state power value at the end of the simulation, the maximum peak value of the current during the entire dynamic process, the maximum voltage overshoot and other key data, which are the scheme simulation data corresponding to this scheme.

[0054] Call the dual-active-bridge power calculation model, input the steady-state voltage value, together with the previously used switching frequency, equivalent inductance, and the three phase-shift angles of this scheme, into the previously established dual-active-bridge power calculation model again, and directly calculate the instantaneous transmission power under the given steady-state voltage and phase-shift angle based on the Fourier series. Cross-verify the simulation results of the dynamic model to ensure that the dynamic model does not deviate due to numerical integration errors or simplified assumptions. For example, there is a scheme in the feedback regulation space, whose external phase-shift angle is 0.205 radians, the primary-side internal phase-shift angle is 0.102 radians, and the secondary-side internal phase-shift angle is 0.051 radians. The steady-state primary-side voltage obtained by the dynamic model simulation is 400.1V, and the secondary-side voltage is 48.0V. Input these voltage values together with the phase-shift angle into the power calculation model. The model performs the summation operation of the Fourier series and calculates that the theoretical power value is 3018W, while the steady-state power directly obtained by the dynamic model simulation is 3015W. The difference between the two is three watts, which is within the acceptable range. Take the 3018W output by the power calculation model as the transmission power simulation parameter of this scheme.

[0055] For each dual-active-bridge feedback regulation scheme in the dual-active-bridge feedback regulation space, compare its corresponding transmission power simulation parameter with the target reference power, calculate the power error, and then judge whether it meets the power error tolerance interval. If the power error is within the power error tolerance interval, the test result of this scheme is qualified. If the simulated power exceeds this range, the test result of this scheme is unqualified. Record this test result for each scheme in the feedback regulation space to form multiple power simulation test results. Power simulation test result: It refers to the judgment conclusion obtained by comparing the transmission power simulation parameter of each scheme with the target reference power, that is, whether the simulated power of this scheme falls within the tolerance interval. The result is usually a boolean value, such as qualified or unqualified.

[0056] The entire dual active bridge feedback regulation space is traversed. For each dual active bridge feedback regulation scheme in the space, its corresponding power simulation test result is read, and the scheme with a qualified test result is searched to obtain the dual active bridge regulation candidate space. If the test result is unqualified, the scheme is directly discarded and not retained. The dual active bridge regulation candidate space is the set of all schemes retained after optimization and screening. Each scheme in this space has been verified by simulation to be able to regulate the output power of the converter to within the allowable range near the target reference power. Through direct feedback regulation, the dual active bridge converter can select the optimal regulation scheme that meets the power error requirements during real-time regulation based on dynamic modeling and simulation regulation results.

[0057] The parameter determination module 14 is used to perform dynamic risk optimization on the dual active bridge adjustment candidate space according to the dual active bridge dynamic accompanying risk mechanism to obtain the dual active bridge adjustment parameters.

[0058] Further details are attached. Figure 2 As shown, the parameter determination module 14 in the precise direct power control device of the dual active bridge converter is further configured as follows: a risk mechanism activation unit 41, configured to activate the dual active bridge dynamic associated risk mechanism, which includes a dual active bridge associated risk analysis layer, a dual active bridge associated risk constraint layer, and a dual active bridge associated risk weight layer; an associated risk constraint optimization unit 42, configured to perform associated risk constraint optimization on the dual active bridge adjustment candidate space according to the dual active bridge associated risk analysis layer and the dual active bridge associated risk constraint layer, to obtain the dual active bridge adjustment optimization space; a risk weighted calculation unit 43, configured to perform risk weighted calculation on the dual active bridge adjustment optimization space according to the dual active bridge associated risk weight layer, to obtain the dynamic risk cost distribution; and a dynamic risk cost iterative optimization unit 44, configured to perform dynamic risk cost iterative optimization on the dual active bridge adjustment optimization space according to the dynamic risk cost distribution, to generate the dual active bridge adjustment parameters.

[0059] Furthermore, the parameter determination module 14 in the precise direct power control device of the dual active bridge converter is also used for: a risk prediction subunit, used to predict the associated risks of the dual active bridge adjustment candidate space according to the associated risk analysis layer of the dual active bridge, and obtain multiple associated risk sequences, wherein the associated risk analysis indicators of the associated risk analysis layer of the dual active bridge include power reverse mutation risk, overshoot oscillation risk and equipment failure risk; a risk verification result generation subunit, used to input the multiple associated risk sequences into the associated risk constraint layer of the dual active bridge, and obtain multiple associated risk verification results, wherein the associated risk constraint layer of the dual active bridge includes power reverse mutation risk constraint, overshoot oscillation risk constraint and equipment failure risk constraint; and a risk verification result screening subunit, used to screen the dual active bridge adjustment candidate space according to the multiple associated risk verification results, and establish the dual active bridge adjustment optimization space.

[0060] Furthermore, the parameter determination module 14 in the precise direct power control device of the dual active bridge converter is also used to: the dual active bridge associated risk weight layer includes power reverse mutation risk weight, overshoot oscillation risk weight and equipment failure risk weight.

[0061] Specifically, the dual active bridge dynamic associated risk mechanism is a dynamic mechanism for risk management of dual active bridge converters. It aims to ensure that the converter does not cause system instability due to sudden failures or fluctuations during operation by monitoring the converter's operating status in real time and predicting risks. The dual active bridge dynamic associated risk mechanism includes a dual active bridge associated risk analysis layer, a dual active bridge associated risk constraint layer, and a dual active bridge associated risk weighting layer.

[0062] The dual active bridge associated risk analysis layer is a machine learning model responsible for extracting and analyzing risk factors from the converter's operating data. It has learned the complex mapping relationship between control parameter changes and transient responses. The dual active bridge associated risk analysis layer is a forward risk assessment model embedded in the converter control system. Its core function is to quickly predict three dynamic risks that the proposed adjustment scheme may cause, namely, power reversal risk, overshoot oscillation risk, and equipment failure risk, based on the current operating conditions and the proposed phase shift angle parameters, without actually performing the adjustment.

[0063] The dual active bridge-associated risk analysis layer employs a multi-layer feedforward neural network architecture, also known as a deep neural network. This network consists of an input layer, four hidden layers, and an output layer, with fully connected layers, meaning each neuron in each layer is connected to all neurons in the next layer. The input layer contains six neurons, corresponding to six input features: primary-side DC voltage, secondary-side DC voltage, load current, outward phase shift angle, primary-side inward phase shift angle, and secondary-side inward phase shift angle. The input layer does not perform any calculations; it only transmits the received data to the first hidden layer. The first hidden layer contains 128 neurons. Each neuron receives the six features from the input layer, multiplies them by their respective weights, sums the results, adds a bias term, and then transforms the result through a non-linear activation function (using a linear rectified function) to produce the neuron's output. The main function of this layer is to capture low-order combinational relationships between the input features. The second hidden layer contains 256 neurons, with a structure similar to the first hidden layer. Each neuron receives 128 output values ​​from the first hidden layer, performs a weighted summation and adds it to the bias, then activates it using a linear rectified function. This layer is responsible for extracting higher-order abstract features, such as the impact of the interaction between voltage and phase shift angle on the transient process. The third hidden layer contains 128 neurons, further reducing and refining the features. Its output represents the deep, implicit features most relevant to risk prediction. The fourth hidden layer contains 64 neurons. As the last layer before the output layer, it synthesizes and organizes the deep features to form the final feature representation used for risk prediction. The output layer contains 3 neurons, corresponding to three output targets. The activation function of these three neurons is a linear function, meaning that the weighted summation result is not nonlinearly compressed, and the predicted value is directly output. The three output values ​​represent the predicted value of power reverse mutation risk, the predicted value of overshoot oscillation risk, and the predicted value of equipment failure risk, respectively.

[0064] In a laboratory environment, a dual active bridge converter hardware platform identical to the target application was built. Using an automated testing system, the operating conditions and control parameters of the dual active bridge converter were randomly varied over a wide range. For each operating condition and parameter combination, the adjustment scheme was actually executed, and the entire dynamic process was recorded using a high-precision oscilloscope and power analyzer. Three label values ​​were extracted from the recorded data: power inverse mutation amplitude, overshoot oscillation comprehensive index, and equipment failure risk index. Each experiment's input features and corresponding output labels were combined to form a sample. The experiments were repeated tens of thousands of times, forming a dataset containing tens of thousands of samples, which was then proportionally divided into training, validation, and test sets. Before training, all connection weights and bias terms in the network were randomly initialized. The initial weight scale was adaptively adjusted according to the number of neurons in each layer, which helps alleviate the gradient vanishing or gradient exploding problems in deep network training. A batch of samples was randomly selected from the training set, such as 128 samples per batch. The six input features of each sample were fed into the network, and the network calculated layer by layer according to the aforementioned hierarchical structure, ultimately obtaining three risk prediction values. The predicted values ​​are compared with the actual recorded label values ​​in the samples to calculate the loss function. The root mean square error (RMSE) loss function is used, which is the sum of the squares of the differences between the three predicted values ​​and the three true values, and then the average is taken. This loss value quantifies the accuracy of the current network prediction. Based on the error calculated by the loss function, the gradient of each weight's contribution to the final error is calculated layer by layer from the output layer using the backpropagation algorithm. Then, the adaptive moment estimation algorithm is used as the optimizer to adjust the value of each weight according to the calculated gradient, in the direction of reducing the loss function. The learning rate is set to 0.001 to control the step size of each weight adjustment. The forward propagation and loss calculation, backpropagation and weight update are repeated, and the network is continuously trained with new batches of data, updating the weights. After each traversal of the entire training set data, the prediction accuracy of the current model is evaluated using validation set data to monitor for overfitting. Training is performed for a total of 200 epochs. After training, the model is finally evaluated using test set data to confirm that its generalization ability meets the requirements. The connection weights and biases between all trained layers are exported as floating-point numbers and burned into the non-volatile memory of the converter control system.

[0065] The relationship between input and output is automatically learned by the neural network during training and embedded in the connection weights, reflecting the physical laws of the dual active bridge converter. For example, when the change in the external phase shift angle in the input is large, the risk of a sudden power reversal in the output often increases; when the load current in the input is close to its rated value, the risk of equipment failure in the output becomes more sensitive to voltage stress; when the combination of the three phase shift angles leads to resonance, the risk of overshoot oscillation in the output increases significantly. The risk analysis layer accompanying the dual active bridge converter, through tens of thousands of training samples, has mastered the mathematical representation of these complex nonlinear relationships, thus enabling it to quickly output the corresponding risk prediction value for a given input.

[0066] The dual active bridge associated risk constraint layer is used to limit associated risks, ensuring that the converter's operation does not exceed safe and stable ranges. By setting different types of risk constraints—power backsurge risk constraints, overshoot oscillation risk constraints, and equipment failure risk constraints—the operating state of the converter is controlled, preventing the dual active bridge converter from entering high-risk regions. The dual active bridge associated risk weight layer is responsible for assigning weights to different types of risks, representing the relative importance of each risk to system stability and safety. A larger weight value indicates a greater impact of the risk on the stability of the dual active bridge converter.

[0067] For each candidate scheme in the dual active bridge regulation candidate space, the three phase shift angle values ​​of that scheme, along with real-time operating data such as the primary-side voltage, secondary-side voltage, and load current, are used to form an input feature vector. This feature vector is then fed into a pre-trained analytical layer machine learning model for forward computation. Through multiple hidden layers and nonlinear activation functions, it ultimately outputs three values. These three output values ​​are recorded and bound to the identifier of the candidate scheme, forming a complete associated risk sequence. After traversing all candidate schemes, multiple associated risk sequences are obtained, equal to the number of candidate schemes. Each sequence quantifies the performance of the corresponding scheme in three types of risks.

[0068] Power reverse mutation risk refers to the risk that the transmitted power will change drastically in the opposite direction of the target value within a very short period of time after switching control parameters, such as the first switching cycle, which may lead to system instability or trigger protection. Overshoot oscillation risk refers to the extent to which the maximum peak value of voltage or current exceeds the steady-state value during dynamic adjustment, as well as the time required for the fluctuation to decay. It is usually expressed by a comprehensive index, such as the voltage overshoot amplitude multiplied by the oscillation decay time. Equipment failure risk refers to the risk that the maximum voltage stress or maximum current stress borne by the switching transistor will approach or exceed its rated limit during adjustment. It is usually expressed as the ratio of stress to rated value.

[0069] For each scheme, its corresponding associated risk sequence is input into the dual active bridge associated risk constraint layer. The predicted value of the scheme's power reverse mutation risk is compared with the power reverse mutation risk threshold of the dual active bridge associated risk constraint layer to determine if it is less than or equal to the threshold. The predicted value of overshoot oscillation risk is compared with the overshoot oscillation risk threshold of the dual active bridge associated risk constraint layer. The predicted value of equipment failure risk is compared with the equipment failure risk threshold of the dual active bridge associated risk constraint layer. Only when all three comparison results are yes, i.e., all risks are within the safety boundary, is the scheme marked as qualified. If any comparison result is no, the scheme is immediately marked as unqualified and excluded in subsequent steps. If marked as qualified, the scheme is retained and moved into a new data structure, which is the dual active bridge adjustment optimization space. If marked as unqualified, the scheme is discarded directly and not included in the optimization space. After the traversal is complete, the original adjustment candidate space is simplified into a smaller but safer optimization space. All solutions in this space simultaneously satisfy the power adjustment and will not cause unacceptable safety risks.

[0070] The risk weighting calculation of the dual active bridge regulation optimization space is performed based on the accompanying risk weighting layer. This layer includes weights for power back-sudden change risk, overshoot oscillation risk, and equipment failure risk. The power back-sudden change risk weight is a coefficient between 0 and 1, used to control the influence of power back-sudden change risk in the overall score. A larger weight indicates a greater tendency to select a scheme with smaller power bounce. The overshoot oscillation risk weight is also a coefficient between 0 and 1, used to control the influence of overshoot oscillation risk in the overall score. A larger weight indicates a greater tendency for the system to select a scheme with smaller voltage and current surges and faster oscillation decay. The equipment failure risk weight is also a coefficient between 0 and 1, used to control the influence of equipment failure risk in the overall score. Since equipment safety is a primary consideration, this weight is usually set to the maximum of the three. For example, a power back-sudden change risk weight of 0.2, an overshoot oscillation risk weight of 0.3, and an equipment failure risk weight of 0.5.

[0071] Each candidate scheme is sequentially extracted from the dual active bridge regulation optimization space. For each scheme, three predicted risk values ​​are read from the associated risk sequence and normalized to obtain the power reverse mutation risk value, overshoot oscillation risk value, and equipment failure risk value. Three preset weight coefficients are read from the weight layer: power reverse mutation risk weight, overshoot oscillation risk weight, and equipment failure risk weight. The sum of these three weights should equal one. Dynamic risk cost = power reverse mutation risk weight × power reverse mutation risk value + overshoot oscillation risk weight × overshoot oscillation risk value + equipment failure risk weight × equipment failure risk value. The entire dual active bridge regulation optimization space is traversed to calculate a corresponding dynamic risk cost for each scheme. After calculation, the dynamic risk costs of all schemes together constitute a dataset, namely the dynamic risk cost distribution. The dynamic risk cost distribution is the distribution of the calculated dynamic risk costs of all schemes in the dual active bridge regulation optimization space on the number axis, reflecting the differences in the comprehensive risk level between different schemes. For example, the associated risk sequence of Scheme A is: power reverse mutation risk 320W, overshoot oscillation risk 150W, and equipment failure risk 0.92. After normalization, the power reverse mutation risk value is 0.64, the overshoot oscillation risk value is 0.75, and the equipment failure risk value is 0.92. Therefore, the dynamic risk cost of Scheme A is calculated as 0.64*0.2 + 0.75*0.3 + 0.92*0.5 = 0.813.

[0072] After obtaining the dynamic risk costs of all candidate solutions, the solution corresponding to the minimum dynamic risk cost is found. Specifically, the dynamic risk cost of the first solution is set as the current minimum value, and its identifier is recorded. The dynamic risk costs of the second, third, and so on up to the 350th solution are retrieved sequentially. For each retrieved cost, it is compared with the currently recorded minimum value. If the newly retrieved cost is less than the current minimum value, the current minimum value is updated with the new cost, and the identifier of the new solution is recorded. If it is greater than or equal to the current minimum value, the current minimum value and the record remain unchanged. After traversing all solutions, the currently recorded minimum value is the global minimum dynamic risk cost, and the corresponding solution is the optimal solution with the lowest overall risk. Based on the recorded solution identifier, the dual active bridge adjustment parameters of the solution are retrieved from the optimization space. These represent the optimal control settings of the converter in actual operation, such as the external phase shift angle, the primary side internal phase shift angle, and the secondary side internal phase shift angle. By calculating and optimizing the dynamic risk costs, the optimal adjustment scheme is accurately identified and selected. This not only improves the stability of the dual active bridge converter under various operating conditions, but also significantly reduces the risks of power reverse surges, overshoot oscillations, and equipment failures, ensuring that the dual active bridge converter can maintain reliable operation while reducing risks.

[0073] In summary, the precise direct power control device for the dual active bridge converter provided in this application has the following technical advantages:

[0074] The signal processing module acquires the real-time electrical signal sequence of the dual active bridge converter and performs sampling interference correction on the real-time electrical signal sequence to obtain a reliable electrical signal sequence. The power calculation module constructs a dual active bridge power calculation model and inputs the reliable electrical signal sequence into the dual active bridge power calculation model to obtain the instantaneous transmission power. The feedback adjustment module compares the instantaneous transmission power with the target reference power to determine the power error signal, and performs direct feedback adjustment on the dual active bridge converter according to the dual active bridge power calculation model and the power error signal to obtain a dual active bridge adjustment candidate space that meets the power error tolerance range. The parameter determination module performs dynamic risk optimization on the dual active bridge adjustment candidate space according to the dual active bridge dynamic accompanying risk mechanism to obtain the dual active bridge adjustment parameters. In other words, by acquiring the real-time electrical signal sequence of the dual active bridge converter and performing interference correction, the instantaneous transmission power is calculated in real time according to the power calculation model. The instantaneous transmission power is compared with the target reference power, and the power error signal is obtained for feedback adjustment. A dynamic risk mechanism of dual active bridge is introduced to dynamically optimize the adjustment candidate space. The adjustment parameters of the converter are optimized according to the real-time risk assessment. While achieving precise power control, dynamic risks are effectively suppressed, and the operating efficiency of the dual active bridge converter is significantly improved.

[0075] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0076] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A precise direct power control device for a dual active bridge converter, characterized in that, include: The signal processing module is used to acquire the real-time electrical signal sequence of the dual active bridge converter and perform sampling interference correction on the real-time electrical signal sequence to obtain a reliable electrical signal sequence. The power calculation module is used to construct a dual active bridge power calculation model and input the reliable electrical signal sequence into the dual active bridge power calculation model to obtain the instantaneous transmission power; The feedback adjustment module is used to compare the instantaneous transmission power with the target reference power, determine the power error signal, and perform direct feedback adjustment on the dual active bridge converter according to the dual active bridge power calculation model and the power error signal to obtain the dual active bridge adjustment candidate space that meets the power error tolerance range. The parameter determination module is used to perform dynamic risk optimization on the dual active bridge adjustment candidate space according to the dual active bridge dynamic accompanying risk mechanism to obtain the dual active bridge adjustment parameters.

2. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The signal processing module includes: The sampling scene parameter acquisition unit is used to acquire sampling scene parameters of the real-time electrical signal sequence and obtain a sampling scene dataset. An interference recognition unit is used to perform interference recognition based on the sampling scene dataset and determine the interference characteristic set of the sampling scene. An interpolation and reconstruction unit is used to interpolate and reconstruct the real-time electrical signal sequence based on the sampling scene interference characteristic set to obtain a reconstructed electrical signal sequence. The signal sequence processing unit is used to perform digital low-pass filtering and timestamp alignment correction on the reconstructed electrical signal sequence to generate the reliable electrical signal sequence.

3. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The power calculation module includes the function expression of the dual active bridge power calculation model, which is: ； Where P is the instantaneous transmission power, V1 is the primary DC voltage, V2 is the secondary DC voltage, f is the switching frequency, L is the total equivalent inductance on both sides of the isolation transformer, d is the outward phase shift angle, d1 is the primary inward phase shift angle, d2 is the secondary inward phase shift angle, and n is the harmonic order.

4. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The feedback adjustment module includes: A control record retrieval unit is used to perform a control record retrieval on the dual active bridge converter according to the target reference power if the power error signal does not meet the power error tolerance range, and obtain a power matching control sample set. A statistical clustering unit is used to perform statistical clustering based on the power matching control sample set to determine the centroids of multiple regulation variables and the cluster dispersion of multiple variables. A boundary determination unit is used to construct multiple variable parameter boundaries based on the multiple adjustment variable center points and the multiple variable clustering dispersion. The parameter feedback adjustment unit is used to perform feedback adjustment on the dual active bridge converter according to the boundaries of the multiple variable parameters, and to obtain the dual active bridge feedback adjustment space. The parameter optimization and identification unit is used to optimize and identify the feedback adjustment space of the dual active bridge according to the dual active bridge power calculation model and the power error tolerance range, and to establish the dual active bridge adjustment candidate space.

5. The precise direct power control device for a dual active bridge converter as described in claim 4, characterized in that, The parameter optimization and identification unit includes: The dynamic modeling subunit is used to perform dynamic modeling based on the dual active bridge converter to obtain the dual active bridge converter model. The simulation adjustment subunit is used to simulate and adjust the dual active bridge transformation model according to each dual active bridge feedback adjustment scheme in the dual active bridge feedback adjustment space, and obtain simulation data of multiple schemes. The scheme simulation subunit is used to input the multiple scheme simulation data into the dual active bridge power calculation model to obtain multiple transmission power simulation parameters; The parameter discrimination subunit is used to determine whether the plurality of transmission power simulation parameters meet the power error tolerance range and to obtain multiple power simulation test results. The result filtering subunit is used to optimize and filter the dual active bridge feedback adjustment space based on the multiple power simulation test results, so as to obtain the dual active bridge adjustment candidate space.

6. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The parameter determination module includes: The risk mechanism activation unit is used to activate the dual active bridge dynamic accompanying risk mechanism, which includes a dual active bridge accompanying risk analysis layer, a dual active bridge accompanying risk constraint layer, and a dual active bridge accompanying risk weight layer. The accompanying risk constraint optimization unit is used to perform accompanying risk constraint optimization on the dual active bridge adjustment candidate space according to the dual active bridge accompanying risk analysis layer and the dual active bridge accompanying risk constraint layer, so as to obtain the dual active bridge adjustment optimization space. The risk-weighted calculation unit is used to perform risk-weighted calculation on the adjustment optimization space of the dual active bridge according to the risk weight layer accompanying the dual active bridge, and obtain the dynamic risk cost distribution. The dynamic risk cost iterative optimization unit is used to perform dynamic risk cost iterative optimization on the dual active bridge adjustment optimization space according to the dynamic risk cost distribution, and generate the dual active bridge adjustment parameters.

7. The precise direct power control device for a dual active bridge converter as described in claim 6, characterized in that, The accompanying risk constraint optimization unit includes: The accompanying risk prediction subunit is used to predict the accompanying risks of the dual active bridge regulation candidate space based on the accompanying risk analysis layer of the dual active bridge, and obtain multiple accompanying risk sequences. The accompanying risk analysis index of the accompanying risk analysis layer of the dual active bridge includes power reverse mutation risk, overshoot oscillation risk and equipment failure risk. The risk test result generation subunit is used to input the multiple associated risk sequences into the dual active bridge associated risk constraint layer to obtain multiple associated risk test results. The dual active bridge associated risk constraint layer includes power reverse mutation risk constraint, overshoot oscillation risk constraint and equipment failure risk constraint. The risk test result screening subunit is used to screen the dual active bridge regulation candidate space based on the multiple accompanying risk test results, and to establish the dual active bridge regulation optimization space.

8. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The dual active bridge converter includes a primary-side full-bridge, a secondary-side full-bridge, and an isolation transformer.

9. The precise direct power control device for a dual active bridge converter as described in claim 1, characterized in that, The feedback adjustment module includes: A signal determination unit is used to determine whether the power error signal meets the power error tolerance range; The instruction generation unit is used to generate a steady-state power maintenance instruction if the power error signal satisfies the power error tolerance range.

10. The precise direct power control device for a dual active bridge converter as described in claim 6, characterized in that, The accompanying risk weight layer of the dual active bridge includes power reverse mutation risk weight, overshoot oscillation risk weight, and equipment failure risk weight.

Images